Web Tools

Search the web and fetch page content.

web_search

Search the web using one of these providers:

• grok (xAI)
• brave
• perplexity
• gemini

{
  "query": "golang concurrency patterns"
}

Output: Search results with titles, URLs, and snippets.

Configuration

{
  "tools": {
    "web": {
      "braveApiKey": "YOUR_BRAVE_API_KEY",
      "search": {
        "enabled": true,
        "provider": "auto",
        "fallbackProviders": [],
        "maxFallbackAttempts": 3,
        "retry": {
          "enabled": true,
          "maxAttemptsPerProvider": 2,
          "baseBackoffMs": 500,
          "maxBackoffMs": 5000
        },
        "providers": {
          "brave": {
            "apiKey": "YOUR_BRAVE_API_KEY"
          },
          "grok": {
            "apiKey": "YOUR_XAI_API_KEY"
          },
          "perplexity": {
            "apiKey": "YOUR_PERPLEXITY_API_KEY",
            "baseUrl": "https://api.perplexity.ai",
            "model": "sonar"
          },
          "gemini": {
            "apiKey": "YOUR_GEMINI_API_KEY",
            "model": "gemini-2.5-flash"
          }
        }
      }
    }
  }
}

Provider selection and fallback

• provider: "auto" uses order: grok, brave, perplexity, gemini.
• fallbackProviders overrides fallback order when set.
• Retryable errors (429, 408, timeouts, 5xx) retry per provider, then can fall back.
• Non-retryable errors (401, 403, invalid request) fail that provider immediately.

Key resolution

For each provider, API key resolution order is:

1. tools.web.search.providers.<provider>.apiKey
2. matching existing llm.providers key (when available)
3. tools.web.braveApiKey (Brave only, compatibility)

GoClaw reads keys from goclaw.json. If you use ${VAR} placeholders there, GoClaw expands them during config load.

Parameters

Example Output

1. Concurrency Patterns in Go - GoLang Docs
   https://go.dev/doc/effective_go#concurrency
   Learn about goroutines, channels, and common concurrency patterns...

2. Go Concurrency Patterns - Google I/O
   https://talks.golang.org/2012/concurrency.slide
   Rob Pike's classic talk on Go concurrency...

web_fetch

Fetch a web page and extract readable text content.

{
  "url": "https://example.com/article"
}

Output: Extracted text content from the page.

Configuration

{
  "tools": {
    "web": {
      "useBrowser": "auto",
      "profile": "default",
      "headless": true
    }
  }
}

Parameters

Limitations

Some sites block automated requests:

• Cloudflare-protected sites may return errors
• Login-required pages won’t work
• Heavy JavaScript sites may have incomplete content

Fallback options:

• Use web_search for snippets
• Use the Browser Tool for JavaScript-rendered pages

See Also

• Browser Tool — Full browser automation
• Tools — Tool overview
• Configuration — Full config reference